Is AI The Scapegoat Employers Use To Explain Technology Layoffs?

AI has not yet replaced actual jobs at scale. It is potentially a cover for cost cutting but might well lead to mass layoffs in the future.

Headlines scream about AI displacing jobs for entry level software engineers. To investigate what lies behind the headlines, I asked a couple of anonymous senior managers in technology firms and a distinguished engineering professor, Garud Iyengar, of Columbia University: is AI responsible for these layoffs? Or is it plain old-supply imbalances caused by other factors?

I deliberately wanted to chat with people on the demand and supply side of the equation. Technologists close to AI tend to almost involuntarily hype demand for the tools they have invested their lifetime studying (Garud is a rare technologist who is pragmatic and balanced). Business executives, at least some, are less swayed by the hype and more level-headed about the costs and benefits of using AI.

The conversation uncovered several nuances that are often missing in press stories.

Tech over-hired during the pandemic

The senior managers suggest that lack of coordination among various divisions, during COVID, led to siloed engineering teams performing the same tasks in a conglomerate. As the dust settled from the pandemic, senior managers took stock of the work being done and realized that several teams performed overlapping tasks that could be consolidated.

Garud counters that these trends are concurrent: “while pandemic over-hiring is real, AI has significantly accelerated redundancy in many tech roles. Tools like GitHub, Copilot and automated code generation platforms have reduced the need for large teams of developers doing routine tasks. The CTO (chief technology officer) of Infosys, an Indian giant, claims that they see a 30% reduction of entry level coders. In the past, overlapping teams might have coexisted, but AI now enables leaner teams to maintain or improve productivity, creating a clear economic incentive to lay off.

But the redundancies would have led to layoffs, regardless of AI. Blaming AI shifts responsibility away from poor strategic planning and mismanagement. It’s easier for a CEO to say “AI is replacing jobs” than to admit: “we miscalculated our growth trajectory.””

Hyping AI works well for tech firms

Simultaneously, tech firms have invested heavily in AI. So, the narrative that AI is omniscient and omnipotent at displacing labor works well to convince investors to pay inflated valuation premiums for companies that expend huge resources on building data centers and hiring AI engineers and data scientists.

Garud adds, “because capital markets reward companies that appear “cutting-edge” and AI-driven, there is a powerful incentive to frame layoffs as part of AI transformation. Many companies are still in early AI experimentation phases—yet layoffs are already being attributed to AI. That’s a red flag. If AI hasn’t been widely deployed in a firm’s workflows yet, how can it be the cause of major workforce reductions? The narrative is being used for optics more than operational truth.

The AI hype doesn’t negate its real impact. Productivity gains from AI adoption are being observed in code generation, customer service (chatbots), and operations (logistics, fraud detection). Tech firms aren’t just inflating valuations—they are seeing real savings and efficiency, making AI a rational business driver, not just PR spin.”

I took this back to a senior executive who countered, “Data in most firms isn’t structured enough for AI.” Another stated, “AI has cut processing time in certain tasks from days to hours. Examples include (i) extracting data and images from various sources on the web; (ii) aggregating such data fast and standardizing them in a usable format such as tables and power point slides; (iii) translating metrics such as from and to the metric system or even condensing technical guidance to understandable business speak; (iv) serving as a great idea starter.

But the work needs constant cross checking when more complex information is involved. Prompts have to be written with extreme precision to stay within context. But there are other tasks where AI makes errors and in areas where the information delivered must be hi-fidelity, such mistakes can be fatal to brand reputation. Even now, chatbots are not good at resolving complex customer problems. They are good at seeking and finding data but fall apart when the customer faces a thorny issue (item has not been delivered, has been stuck with customs for days etc.). Often, the contact is sent to a human agent after conversing with a chatbot.”

On balance, I wonder whether the cash flow savings can ever justify the inflated valuations we see today. Perhaps growth in such savings or new revenue streams might become a large cash flow stream but a lot of the valuation may be faith-based.

Excess supply of computer science graduates

College graduates have been told for years that coding is the ticket to an upper middle-class life. The number of students graduating in computer science has more than doubled over the last 10 years in the US and Canada. These numbers are even higher overseas. Even non-computer science majors get a lot of computer science exposure. Inevitably, supply exceeds demand, even if there were no AI, at some point.

Garud states, “we are no doubt seeing a market correction where supply is catching up with demand. It’s much more palatable for companies to say, “AI replaced you” than to say, “we no longer need as many engineers because the market is flooded.” AI becomes a psychologically easier rationale for both internal morale and external messaging.

While oversupply does exist, demand is also dropping faster because of AI efficiencies. AI doesn’t just reduce the number of required coders—it changes the nature of the work. Fewer engineers are needed to deliver the same or better output, especially for front-end/back-end tasks that can now be scaffolded automatically. The presence of excess supply doesn’t preclude AI being a causal factor in fewer job openings.”

AI potentially masks continuous outsourcing?

Senior executives I talked with suggested that computing is a job that is easier to parse out into smaller sub-projects and hence is more friendly to remote work. Overseas software engineers are significantly cheaper than local graduates. And, we are not talking about Bangalore, India. An engineer based in the UK or Europe costs significantly less than an engineer based in NYC or the Bay Area.

Garud adds, “firms have strong economic incentives to replace expensive domestic labor with cheaper, equally skilled overseas workers, a practice that predates the AI boom. Citing AI masks the continuing globalization of tech labor, allowing firms to make cuts while maintaining a “futuristic” cover story.

Offshoring is a long-standing trend. What’s new is that AI reduces the need for offshoring too. LLM-based (large language models) tools can generate documentation, translate codebases, and offer tech support—all tasks previously offshored. In fact, AI is displacing both domestic and offshore workers, making it a direct contributor to the job squeeze across geographies. A case in point are the layoffs in TCS in India.”

Entering graduates take longer to train

Senior managers suggest that onboarding new graduates is expensive and takes a year or more. Firms were willing to incur resources to bring new graduates up the learning curve when software labor was scarce. They are not as interested in doing so anymore because they can easily get seasoned engineers today at the same or a lower wage than what was paid to entering graduates a couple of years back.

A senior executive states, “as long as a student did something aligned with computer science, the student would be trained as a software developer and land a six-figure job. No longer. Now, you have to specialize more and get an advanced degree to land an entry level job. This is also partly because universities do not prepare students for the tasks we look for when we hire fresh graduates. They take significant retraining which, all else constant, we look to minimize or even eliminate.”

Garud suggests that “companies possibly prefer plug-and-play hires, not because AI is so advanced, but because financial discipline has become stricter post-2022. Again, AI is the cover for cost-cutting.

In the longer term, AI may make onboarding juniors less necessary, not just less desirable. Many entry-level tasks have become automatable—documentation, bug fixing, testing—undermining the value proposition of junior hires. In other words, AI has changed the ROI (return on investment) equation on entry-level hiring, reinforcing its direct impact on layoffs. There is an inherent paradox here –the entry level tasks can serve as low stakes “learning by doing” tasks, and with these tasks gone, it is not clear how to get the workers the intangible skills they need for training and eventual advancement.”

The market clearing wage now is not $165,000

All the factors discussed above lead to a lower market clearing wage for tech workers. But wage expectations are still pegged to older numbers.

Graduates now know what their peers in big tech earn. Accepting a wage that is significantly less than what a peer might make at a big tech firm is a bitter ego pill to swallow although the market clearing wage now for their skill level is only $75,000 at a mainstream firm. New graduates feel that they are underpaid and are hence perpetually dissatisfied. No employer wants to hire someone who is perpetually unhappy about being with the firm.

Garud counters, “while expectations may rise due to transparency, this alone may not be able to explain widespread layoffs. Dissatisfied workers can be managed through culture and HR. Layoffs are financial decisions. Firms may be choosing AI tools over junior hires because they reduce onboarding and payroll costs—not because workers are too picky.

Rather than admit that they’re unwilling to meet higher salary expectations, firms find it easier to say: “we don’t need as many people thanks to AI.” In reality, they don’t want to pay what the new labor market demands and that demand itself may not reflect the new market clearing wage.”

Hesitation to invest because of policy uncertainty

Companies have placed investments in capital and labor on hold till the dust settles on what US trade policy looks like. Most investments take five to seven years to pan out, if they ever do pan out. Lack of predictability about what duties, tariffs and trade policy will look like over that period leads to a wait and watch response in boardrooms. Hence, hiring, other than of the AI sort, is on hold.

It’s not all doom and gloom

An executive points out that the supply glut is not necessarily all that negative. Ageism has always been a problem in tech firms partly because there is an incessant supply of new workers well versed in shiny new tools. Hiring seasoned workers may help address the bias against older workers in tech firms to some extent.

On top of that, under-employed software engineers are more likely than others to venture out on their own and create start-ups. This is a risky endeavor for sure. But one of the new 100 startups that the surplus engineers create may become the next OpenAI. Mainstream non-tech firms can also pick up talented tech workers at a reasonable cost.

This is not to say that AI will not lead to layoffs in the future. All I argue that is the current narrative is arguably overblown. CEOs need to villainize someone or something to justify layoffs to the remaining workers in the firm. AI serves as the perfect excuse for now.

I’ll let Garud have the last word: “I agree that for many firms, AI has not yet replaced actual workflows or jobs—at least not at scale. Instead, it is used symbolically: to impress investors, manage employee morale, and justify corporate restructuring. The perception of AI outpaces its practical deployment.

AI becomes the perfect non-human scapegoat: it’s inevitable, it doesn’t sue, it doesn’t organize, and it doesn’t tarnish the brand. CEOs can point to AI as an “external force of change,” instead of taking accountability for strategic missteps, bad acquisitions, or investor appeasement. Just as “globalization” and “automation” were buzzwords used to justify offshoring and restructuring in previous decades, “AI” has become today’s universal rationale. But many of the forces driving layoffs – overcapacity, labor arbitrage, cost control, macroeconomic hesitation – are neither new nor AI-specific.

AI may be a convenient explanation—but it’s not an inaccurate one. The real economic pressure from AI adoption is already being felt across industries. CEOs may be citing AI not merely deflecting blame, they are likely pointing to a genuine shift in labor needs brought on by new capabilities. Unless we plan ahead, AI might truly become the villain.”